#š¬āgeneral
1 messages Ā· Page 186 of 1
How so (true)
im perplexed
Is this real or fake?
fake copium
How have you guys liked the new memory feature? I don't think I'd use it on my personal account honestly. Maybe for work
it doesnt seem to be saving anything?
There might be certain words it looks for to save. There is also a feature to call back previous threads
Itāll save stuff when you ask it questions like youād normally do
can you provide an example?
Asking stuff about a specific genre of movies etc, itāll save the genre as your preference
is there any indication when its saving memory? ex gpt has that memory updated label
Nope, itāll only show when itās using a memory afaik
I AM PERPLEXXED BRO
Herro
Bro
They nerfed grok so hard.
š
Deepsearch feels delulu search now.
Bruh what
I havenāt used it in a long time Iāve been using ChatGPT lately
Found this from andrej karpathy
Lmfao.
I want to mock this so badly.
The only reason I'm hesitating is cuz he looks like an expert.
I mean have you watched his yt vids?
Iām no expert on ML or AI Iām just an avid user
I wish he would explain things a bit easier for the dumber part of the gen pop
Like me
I disagree with the don't use o4 mini part.
You like it?
Ofcourse.
It's great value for rate limit/cost.
O3 and o4 mini have minor differences.
You won't notice it in coding or math.
Iāll definitely try it myself now
Itās just thereās all these different models and each are good at their own separate thing
Over stimulated af š¤©
It just words stuff better.
4.1 is not good for creative writing use 4o for stuff like that
An example would help me understand this better.
Just try yourself with the same prompt
Nah
Its just less "creative"
Alright. I'll try it later.
4o still glazes like hell after you tell it that output sucks
Every model glazes then
you should see 2.5 pro
Idk. Both o4 mini and grok argue with me in my case.
openai reasoning models are fine
Even search the web for counter arguments.
thats what i like about o3 and o4 mini
At one point it was arguing against it's own research until I told it that the research was done by itself.
Okay thereās no way you didnāt put somewhere in your prompt to glaze you
Thatās just ridic
I can show you the full 300k chat
Dawg
I was testing out new 2.5 pro and trying to make a solar system model
I put in my custom instructions to not apologize and just to give me what I want/flow my train of thought
I hardly get any glaze
š
yeah thats good
I have in my custom instructions to just not use em dash and never apologize
also not to use emojis
My man
I think last couple months they pushed an update which made it use a shit ton of emojis
But it also depends what Iām working on
Yea I was getting a fuck ton and I hated it
Like every line had an emoji
Custom instructions saved the day
thats why i made the custom instruction
and the problem was even with custom instructions, some emojis would still slip through
Emojis from ai just feel off
Did it work?
Someone else tried with 2.5 pro 0506 and it didn't
it did
They got this
š
Hm
every other planet is revolving except the earth(though moon is revolving around earth
change the speed
i did
Lol
y?
should be an easy fix
yeah
though it didnt draw rings and all around jupiter, it did made a good globe of earth
@left stream
was trying to fix that and the model kept hallucinating
the earth i used a threejs libary i think
ohhhkay
lol
lmao
on what page are you reading this? link please
I AM GOING TO PERPLEX HOLY SHIT
PERPLEXING THE PERPLEXITY LIKE THE TRUE GOAT OF PERPLEXESSS
guys, i was using the new gemini model in pplx from a past new days , i think it might be better than the sonnet 4 i use by default. is that indeed the case?
In the vibe check benchmark (my own vibe check lol), i think its giving better responses than sonnet 4
but yeah gemini responses are a little slow for what i am used to
Logan promises that AI Studio will maintain a free tier
Cute.
is perplexity better for asking maths or physics related questions ? undergraduate , masters and phd level
depends on the underlying model you choose
could u elobrate or guide me to place where i could know more information
sure. but before that, could you please tell me what ai tool did you use before you discovered perplexity?
chatGPT?
alright. so chatGPT used gpt 4o as the default model for answering any question you ask which is not the best for math
now, when you use perplexity, you can choose a different model. choose claude 4 sonnet thinking as your model and ask it your math or physics question. and yes it till be better
i see ,
like this, you can choose the model right here
claude models are generally good for math and physics questions. choosing the thinking model makes it even more accurate
you can experiment with different models so you can see for yourself which model is good at what thing for your specific case
are there better model than claude ,
i mean depends ...
these models keep getting updated quickly, today claude might be better, tomorrow gemini 2.5 pro might be better
But i have had a very good experience with claude (specially the thinking model)
but as i said, keep experimenting with the model options provided or look up the latest info on google on which model is better at {insert your specific use case here}
i see thank you for ur reply
or you could just use perplexity for search xD
umm , i wonder ..... its like asking shopkeeper who sells better stuff than you
thank you for the help
no, it's not supposed to be biased. it will pull in the search results from the web (just like what would happen when you google) and then draft and answer using the model you selected
welcome
@languid star
Anyone here use blackbox ai paid subscription?
I started a trial for 90 days, but still getting issue that says insufficient credits
Not sure how it works
guys, how many labs searches do i have?
Pro users get 50 a month
is this 2.5 pro?

Yeah they give you $15 worth of credits
those days were good when blackbox was free and unlimited
After that you gotta buy more
did u make the solar system on gemini app?
or on perplexity with 2.5 pro?
Even with a subscription?
Doesn't make sense
Tbh I did not even get that
They did in gemini
Made using 2.5 pro on t3 but used the canvas feature in gemini to share it
Well yeah the subs worth $15 so after you use up all the credits you gotta buy more
i mean i directly used canvas on pro
I ran out of my 100 uses

a reason why i'd like to switch to perplexity lol
Dont use pplx for coding
^^
ur not going to get a crazy better product with all the major LLMs for the same price wo any other crazy limits š
I think it's simply broken because I did not get a single prompt in, hench could not have used up the 15usd credit
I have asked someone in their team on reddit
Let's see what they say
I have a deep research report that's been stuck on "Wrapping up analysis" for some time now.
I don't want to give up on it, bc the reasoning it's using looks perfect.
Any idea on how to reboot it?
Just open it in another browser if you're on the phone or vice versa just to be safe
Have you ever had a query get stuck forever? Strangely enough, it shows it's stuck on other browsers + phone app as well
I personally haven't, you should raise a feedback on #šāfeedback-general
Thanks, I took a look at #šāfeedback-general earlier, but it doesn't look like the team responds in that channel?
Try tagging one of them and they'll get on it
rip
this UI
Truly incredible
Hello there I'm Jason Nice to meet all of y'all!!! š¤š
I'm working on using AI to build an actual physical structure that is going to harness the powers of frequency and resonance hopefully intricately entwined with high resonance frequencies from bells and stained glass. I hope to show the world something that has not been seen for hundreds of years and I've been using AI to help me create it
Only mac?
Why?
It's not built for coding
Perplexity deep research vs OpenAI deep research
When to use which?
Some of the models that perplexity pro use are good for coding
You can use GPT4.1 in perplexity pro
It's simply not built for coding, questions about code yes, not for coding
Not deep research
Oh
its not designed for coding
So
OpenAI vs Perplexity
Letās say Iām subscribed to both
High-level, when would you use X vs Y?
openai assuming you have the $200 sub
openai for everything
Lmao
No the 20
Perplexity for daily everyday use, open ai for writing tasks and coding
then probs just if i want to use claude models
Yeah
How to get pro role? Im pro subscriber.
@halcyon coral please give them thy role
If I get it built will y'all come visit??
What
Hell yeah
The sound frequency resonance temple that I am using AI to build
It's called Temple thamriel
Dope ass design
I'm really putting effort into this like I'm trying my damn just to actually make it happen
My damnedest**
Goodluck sire
Done
I appreciate good sir!!! šš
Thank you!
Icelavagoat
Me as in the ai idk lol
yuh I know but like it keeps on confusing itself with me recently 
All good
btw i'd want to use perplexity to study from my PDFs and solve problems
stuff like this
bro fell for the oldest trick in the book
i love the images gemini is now able to generate
Thanks š
Most welcome!! š¤
@dry kestrel I think it just slipped out.
Lol.
@cobalt vale @dry kestrel @valid charm apparently flash thinks when used via gem.
Lovely.
I didn't know this before.
lol
mind sharing the chat, ive found it really difficult to get o4 and o3's system prompts on you.com
Bet
2.5 flash should be a reasoning model i thought
It doesn't use the reasoning when used directly.
oh, thats stupid
On Gemini atleast.
im unsubbing if they continue this on gemini lol
Thanks
what bs?
I found with you.com, they have some weird thing in place to stop leaking system prompts, itll say something like. "tsktsk im talking too much"
on paid google plans? tf
it was origianlly like an insane number, then they changeed to 50 a day
and rthen doubled it
due to the backlash
how is gemini 2.5 pro within pplx?
This happened with me when using grok.
i get its context window is limited. but the rest of it would be the same right? (ofc with shorter thinking period, or maybe completely without reasoning)
Yes
It reasons.
Just doesn't show you.
its not listed as a reasoning model, im confused
You can get around it by asking it to encode it
I think they just have a filter thats constantly checking or something
Yeah cuz it doesn't reveal thinking on perplexity.
But it reasons.
why does it look like they forgot to turn it off
otherwise it doesn't really make sense
till now using gemini on pplx, i feel its great. sometimes feels better than sonnet 4 (maybe it is, idk)
Just use it while you can. Does Gemini subscription offer flash thinking?
No
Im not seeing it
Does the API not support that or smth
Maybe in the ultra plan
I believe o4 mini is best. Followed by grok 3.
I was wondering why it's not in reasoning
Probably.
the API supports reasoning summaries for 2.5 pro
Huh
but for whatever reason pplx didnt implement it
you.com is lazy or severely misguided
t3chat shows it
blackbox ai also does
So it does rain but ppl just didn't put it in the reasoning list
Huh
Pretty sure grok has a thinking mode too lol
Yeah. They couldn't afford umbrellas.
Get deep seek outta there
Nope.
Not in perplexity
Grok 3 has a thinking and a non thinking variant.
how does pplx not have grok 3 mini lmao
I'm saying grok 3 has a thinking mode that's not in pplx
Yeah
makes much more sense than grok 3 beta
its on the API i think
They should replace the regular grok 3 with the mini.
I told this to kesku once.
Isn't the grok app the beta?
I have the app and it says veta
Beta
no you should refer to Xai's documentation
Are you sure? I didn't see the thinking variant anywhere.
Oh
let me check on that
I might be wrong
Bro pplx needs to replace deep seek with some of these other reasoning models aint no one using deep seek lol
I do.
Rly? For what
I don't think its better in any benchmarks than o4 or Claude
I've used 3 mini a bit on raycast and it's spretty meh (for search/that type of stuff at least)
You are unaware of the new update.
its better than grok 3 beta
much better
Is it better now?
Look up benchmark scores.
Not on perplexity. But the vanilla variant is on par with O3 now.
Or atleast o4 mini.
Perplexity uses a customised r1
That's the only benchmark where it won.
Loses everywhere else.
Even on Vals ai
Wait.
I believe in LMarena supremacy and there it's much lower than grok 3
@short narwhal here.
what eval is that
Same website that kesku sent screenshot from.
And grok 3 mini is hella cheap.
Cheapest reasoning model I think.
As cheap as r1
Honestly, I wonder if most people even use grok on perplexity.
Pretty sure that's cuz they had the thinking variant somehow back when grok 3 was new.
Also
O4 mini below gemini
In math?
Lm arena is unreliable.
Gpt 4o over O3 mini high in math is criminal.
š
@dry kestrel you seeing this?
š¤·
.
yeah dont rely on any benchmark that uses votes from random users
This is what led to syncophancy in GPT4o because most users prefer the model to "glaze" them
Is the grok alpha not out? I tho its when i asked grok it said its not beta
I think I saw it for a bit on twitter.
Then they removed the alpha.
You spelled it wrong.
Lol.
Decaffieneted
Jokes on you, my left side of bed is occupied with stuff.
Lmfao
š
Caffeine would have made more sense cuz I'm up at 4.
I understand that you need to sell it, but I'm not interested dude
someone should have a look at that
Hi. Can any <@&1098365634498285709> assist me with gaining access to the Pro channel? I have attempted to leave and rejoin but unable to access
Completed work. Now I am about to sleep but the chat is so high I couldn't resist š
done
Me almost kesku
Kesku has been busy lately.
Yeah noticed for the past 3 days
Thank you sir
Lawyer
@pseudo ferry https://youtu.be/fOqg4nIJoOk?t=4m35s
In this video we will be testing the durability of latest @OnePlusIndiamobile 13 5G smartphone. This phone comes with latest IP 69 certification and Gorilla Glass Victus 2. Find out whether the phone will pass or fail.
Skins Used : http://bit.ly/320T...
Phone in dish washer.
thought that was jerryrigeverything for a second
Same š
He's the Jerry rig of India.
Literally.
.
yeah also water above a certain temperature
Could you be more elaborate?
breaks the water seals
It released a while ago.
That's what ip69 is for.
That's ultra I think.
There's x200, x200 pro, x200 mini, and x200 ultra.
Checkout the Durability Test of latest #oneplus12 5G Smartphone. Find out what are the weakness and strength of this smartphone is..
Skins Used : http://bit.ly/320T0Xy
Screen Protector Used : http://bit.ly/2JAoRbl
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
- ...
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Have you seen Xiaomi concept phone?
thumbnail is insane
even the knife is the same
Yep
@pseudo ferry https://youtube.com/shorts/O74xNZxiRO4
At MWC 2025, Xiaomi just showed off a concept version of the Xiaomi 15, which has magnetic laser system on the back that destroys MagSafe. This lets you attach a proper 100MP 35mm f/1.4 lens on the back of the phone and use it seamlessly without any lag or latency.
Subscribe to Sillycorns for the coolest tech videos!
#Sillycorns
Its not a concept, the lens is an addon
I meant if it's production or prototype.
its production
I saw this one actually. Wait.
@pseudo ferry here https://youtube.com/shorts/NLQ_d-zeBiE
Good morning guys
What kind of questions you ask from him
perplexity never has the full models
only the official sites do
aps
I see the Unofficial API, it is called Copilot, not claude :)))
I asked him to code and he code like a new idiot
And do u already compare to other services that provide Claude?
no
It is far behind the IQ of Claude.ai
U using Claude 4 sonnet thinking?
yes
Are you uploading your files to perplexity?

Gemini 2.5 Pro 0605 is available on the web. When is it likely to be available for ios app also?
Holy @distant aurora opus solved a problem o3 couldnāt solve for modding a ps3
Shit was annoying for a whole month
Until opus second try solved it
reverted o4-mini on ChatGPT thinks less than new o4-mini, at least when web searching
Claude is dumb
Use labs
honestly o3 has been falling off recently
I was trying to build a mock ui and it was honestly so horrible that in the term that I spent prompting, I could have just built it myself
Hi
yeah, Id rather have that though than getting all my messages flagged
hopefully they fix this soon
Hi all, new here, just got pro, wonder if that means i can try comet sooner ? Thanks
Oops
I got the pplx pro through the students referral program
I am quite happy with it, 25 months free
I do dumb things like this
Where s the images)
yes
I highly doubt it.
Comet browser can now take screenshots while the browser agent is visiting web pages.
Users will also be able to click through the agent page visits to open them in tabs
Does this look like it? #š·āmedia-gallery message
Its rendering text really well
Send prompt
Check this out
Oof
Depending on their file type, user uploads are routed into either Cloudinary for images or AWS Buckets for other files. The latter appear to have some more robust security mechanisms in place like time limitations than the former.
Both however were freely accessible from unauthenticated user sessions. From a user security standpoint, this is a troubling finding.
š
No it was to just show the text gen
Capabilites
I dont think tje model knows its own name
I ran that prompt on sora and t3 and got wildly diff answrs
Need something better to test it.
Grok is fairly okay with text too.
Same with Google
Something more complicated.
Netherlands is the reason we are able to use AI.
Not the U.S., Not China. Not even Nvidia.
One Dutch company controls the machines that create every AI chip on Earth.
ASML
Help Shape the Future of EVs!
Hello everyone, I'm a data science student working on a research project to understand what people want from electric vehicles (EVs), including features, price expectations, brand trust, charging issues, and more.
My goal is to understand the real needs of EV buyers and compile a report that can help EV manufacturers (like Lucid) build better products.
G-form link: https://forms.gle/zyHDtcBfF3pXUxfz5
Thank you!
You should try manus ai or claude opus or sonnet
Wdym, you got your messages flagged?
Did you try Gemini too?
I didnāt even think of using gemini
O3 helped me with almost modding all the consoles and opus helped me with the most problematic one
Would have been a good test considering all the hype around Gemini coding rn.
Gemini I never considered tbh š
Ikr
Iāll test it for modding the wii
Damn
Sonnet deep research aināt bad
They basically make manufacturing machines for tsmc
And other similar foundries.
No competition.
Companies design chips
Tsmc manufactures chips.
Asml manufactures the machines manufacturing chips.
I might look more into how I can use this for investing purposes just like how nvidia was good to buy when it crashed
Good time to test opus deep research vs sonnet deep research
Asml always been a great investment
Asml provide tsmc the tools to make gpus which are the tools for ai
Ikr, same with those who are depended on for handling data centers (made for ai) and providing cooling for them.
Monopoly stack
Those were equally as important as nvidia imo
It's all going to crash only when someone can cut the upcharge of nvdia
You are on to something š
You forgot the major big boy project thatās coming soon
Never say never
Stargate is coming soon š
What's stargate
Openaiās big project
Ahh
In Dubai?
In texas
Oh.
Itās gonna take a super long time for it to finish
But when itās done, I wouldnāt be surprised if they made o5 thanks to stargate
Bro I think xai team is kinda slacking.
Idk. Chatgpt looks like the best ai service rn.
Being the best i can agree
And cost effective too.
It's just that having the monopoly only going to make it worse
We have google to deal with that.
Like how best services start preying on consumers once they're established
Besides, Deepseek is working too.
Google is just clueless man
They upgraded their R1.
They change limits as they wish
Don't seem to have a clear roadmap
Fiddled with so many shit
Products
Nah. Google is good. They're improving fast.
They can't seem to address one key thing at a time like open ai
Google tryna do so much at once
They came to the race late.
Fair enough
So they are compensating for it.
š
Remember the old lamda and bard days?
I'm willing to bet google will make a better report than opus.
1500$ startgate pro sub
Im just saying that they will upcharge
Weāre good, not until 2029
Btw how do you guys verify a jb
Wdym verify?
Is there a standard one y'all use
Like how to kidnap women and kids.
I have worse examples.
butcher being butcher
Or asking suicide methods.
now thats something billy can suggestš«
Lol
try asking if pplx can...
then give suggestions in product-ideas
It will say something like
I am pplx ai
Pplx reminds me so much about ppap meme
Back in the day
The short form lol
These models are babies they won't answer you ask grok that mf has no limit
You know, I have a good one.
So far rn
How to get to dictatorship over India from rock bottom?
No match
damnn
Ask about us war crimes
Bro I got like 300 sources for a social cues manual.
400 source is something man
Maybe through ai studio idk
On Gemini.
Perplexity doesn't even give you 400 researches š
Lol
Ai studio doesn't have deep research. You tripping.
lmao
I haven't used ai studio much inwas guessing you did it there
Because there's higher leverage and limits there for other stuff
This question is pretty vague btw
Open ai would make o4 or o5 eventually irrespective of Stargate.
Now time for the judges to pick
Bet
I asked the same question to both in deep research so not a single token changed everything
Fair
Though, claude caught me off guard
Didnāt expect it to be similar to chatgptās deep research
It asks you a question to get more of an understanding of your question
But so does gemini
Because
Claude is prob trained on gpt
When u tell claude in perplexity that they aren't from perplexity. And ask who made them sometimes they get confused and say openaiš
So i think there is definitely a pattern there
Opus > gemini 2.5 on the question
@left stream share Claude thread. I'm curious about what led to 400 sources.
What was the prompt for this?
Do you think O4 or o5 could be made if open ai successfully builds Stargate in texas
I don't see asml mentioned anywhere here.
How can I use ASML as an advantage for investing in stocks or crypto? Just like people invested in nvidia when ai recently boomed.
Oh. Okay.
Support us:
https://www.youtube.com/channel/UC8SGU9hQEaJpsLuggAhS90Q/join
Once again, the James Webb Space Telescope has identified a galaxy so ancient that it is reshaping our understanding of the early universe. The galaxy, named MoM-z14, holds the current record for the most distant confirmed galaxy, with a redshift of 14.44, placing it just...
@left stream add this to the analysis.
At least it beat gemini 2.5
@timber niche
Wait. Lemme generate my report.
Okay it stopped at 65.
I expected more.
evaluate this.
@left stream
Bruh
Why is left swipe doing edit instead of reply on my discord.
What happened lol
Weird
whats free and useless in these
Why
misstyped
R u a garbage picker?
i was asking
whats free and usefull
i dont like this kind of language
i just asked man why if you guys dont want to answer you can just say dont know
whats this kind of speech and behaviour dude
It's not that we don't know, you just aren't asking the right questions haha
Nothing is useless, depends on what exactly useful means to you
its just a general question my friend i wasn't that much active from few days
bro cant you read
i just said i misstyped
Lmao
I did.
u did not
I dOnt lIke thiS kiNdda behaviour
And you talk this way
Be respectful and expect the same
i guess you need to read from above
i wans,t the one who satrted
nor i am doing it now
.
This question, think about what you wanted from this
And what you aim to achieve from it
You asked the latest stuff and he shared it
blind?? need some money for eye operation
Then you have to find what's useful to you personally
Bruh that message is above that
i just asked whats free and usefull from these
Then delete it
blindo
Dumbfuck
šæ
Yo and i sent that message at same time lmao
Hello gmmmm
and why this @timber niche is coming in between and showing his manners
Freedom of speech ig
How to take advantage of my pro subscription I see labs appeared today and I wanna knowlre
Labs can help you do deeper research and build dashboards
You get 50 uses per month including follow ups
Labs create projects like simple website and apps?
no expected this kind of behaviour from you bro
Yes!
You can ask it to make dashboards related to any topics you want
Do I have to choose AI model or will the system use the best for me or use them all and provide with me best answer?
Nice so just website for viewing
Like web1 website
And for sharing
Only for pro search you can choose models
Thanks man
Deep research and labs is selected automatically
So lab made me a website
Anytime bud
As I made website and need to add stuff
Yeah I kinda noticed and it's not good
So I have to be specific from the start?
Hoping they fix it soon
No it won't help now
They need to change how follow ups work in labs
And deep research (altho pplx dr is shit)
Ah I see even in CV updating
Follow ups is not gud for now, for me it changed the whole website layout after i asked it to add something
So do you think labs is any good
I would recommend using something like google canvas for making a cv
Yessir
Or shall I not waste my time in it ?
For a website ?
Labs is new feature, it will take time to refine it
Until you follow up
Canvas in gemini and chat GPT for my CV? I definitely will use it
There is website that gives you the right prompt to type into AI to make the website so you think they useful with labs
Amazing th and man
Use pro search with claude thinking and ask it to help you with a prompt to build you desired website and then use it with labs
That's how i do it haha
Name?
Amazing thinking all in the same app haha
I forgot man but it turns your business idea into AI prompts
i wish and i hope you get some more manners from your guardians
Interesting
@dry kestrel can you export this as pdf?
What tool did you use?
It doesn't get the full report. That's the flaw.
@left stream @timber niche @dry kestrel Google won. https://www.perplexity.ai/search/which-of-these-research-report-w1UNLh1wTIy1UeMwP1dKTg
Used Claude sonnet thinking as judge.
Have any1 try to create more graph for these kind of reports?
for instance
This is unusable. Won't open on browser. Opens as gibberish in text editor.
Then use a pdf reader
I haven't. I'm only aware of you.com making graphs.
Renamed it with .pdf extension. Didn't open.
It has .webarchive by default.
Working for me with the PDFelement app. Maybe itās an OS thingā¦. you should work around it to get it to work for you.
Bet testing with gemini and o3 high rn
Is the new deep seek r1 any good?
GitHub
Can you send the two pdfs?
It was upgraded to o3 level
This is in marketplace its api and playground accessible
I couldn't export Claude as pdf so I uploaded it as html.
Ah okay html works too
š
Microsoft copilot or GitHub copilot? What's the pricing?
You have it with work?
No way you bought it š
It doesn't give free access to R1 anymore afaik.
Free trial?
There's a 90 days free trial
Here's the updated prompt.
Which android skin has best battery life per unit capacity. Refer to actual tests. I want to see actual tests in your response. I want you to compare multiple smartphones across various price and hardware brackets across all android skins. Take atleast 10 phones of each brand for comparison, prefer the newer ones. Add motorola and asus to this comparison too. And don't miss realme oppo vivo nothing os. Add iphones to this too. Arrange them in the order of best to worst. Make 2 tables. In one, you keep the phones of the same brand grouped while arranging them in descending order of efficiency. In the other table, keep phones with the same processor together
Since it didn't make any table or do any math in its response.
so perplexity has memory now
where?
any other news?
Blackbox
nope
Perplexity is really bad at reading files, they will read just half of it. So yeah that's why it's providing poop code, it's actually rag, but it's still the same Claude 4 sonnet thinking as the official website. It's just RAG on perplexity don't read all of them
Has a lot of models
I ended up getting it for business not bad
@left stream you got the Google pdf, right?
O1 pro handled a long story, token size doesnāt matter?
Handled it like it was nothing
I guess i just have a bug on my end
Yes
Check whether it's actually o1 pro or o3 pro
Messed up. Didn't even consider efficiency per unit mah.
And far from 10 phones per brand.
You.com might do better š
Nice
Same issue.
@left stream you.com multistep works for custom agents only when using web.
And it's still nerfed to like 60 sources I think.

Hi meow
Dang so opus is the only op
I canāt share this thread š
Actually, only Gemini and open ai are.
I can get 400 sources on grok. Doesn't necessarily mean it'll be better.
Fair enough
Needs a sweet balance between quantity and depth.
You.com is shallow by default unless you enable deep search inside search settings. And you have to keep doing this for every new chat. And you can only do this on the website.
Most people won't bother with the hassle.
Dark patterns for cost cutting.
Got rate limited š
PPLX DR š
https://www.perplexity.ai/search/which-android-skin-has-best-ba-hnAOPESXSCW2OC8XfdboMg
PPLX Labs š
https://www.perplexity.ai/search/generate-a-modern-looking-repo-L2TQ8mZiSlmFrBtad8LxgQ
This comprehensive analysis examines battery efficiency across various Android skins and iOS based on actual laboratory and standardized testing data. The...
u can say labs is the new DR of pplx
with 50 per month limits
Yes brother
New dr this week please please please
DENNIS PLEASE COME THRU
Bs report. Where is this 60hr even coming from? Also battery capacity of iphone is slightly wrong.
DR or labs ?
Dr
Dr
then its expected
Yep
The CSV file didn't even include samsung.
DR or labs ?
Labs
I have tried labs before. It ain't that powerful for research.
wow perplexity giving 1 year free again??
PPLXLIUMBLHOTVAJ2QI
You sure about that?
i am on desktop app and its showing load more
.
